TECHNICAL FIELD
[0001] Embodiments described herein relate generally to radio frequency (RF) communication
devices, and, more particularly, to the generation of RF clock signals by an RF communication
device.
BACKGROUND
[0002] In RF communication systems, clock signals may be used for a number of different
purposes. RF communication systems require good clock signals to work properly.
[0003] In an RF mobile terminal, RF clock signals may be created on-chip by phase-locked-loops
(PLLs) locking at an external crystal resonator as a reference clock input with different
dividing factors. For transmitter and receiver circuits running at different frequencies,
normally at least two PLLs are needed in an RF integrated circuit (IC). For some mobile
terminals, such as mobile terminals that support multiple RF communication standards
(e.g., WLAN, Bluetooth, GPS, etc.) additional RF clock signals, and hence additional
PLLs, may be needed. Further, in some suggested radio standards, in order to increase
communication data rate, carrier aggregation is suggested, in which additional RF
clock signals may be required.
[0004] Creating many RF clocks using PLLs, however, is not an easy task. When several PLLs
are integrated into a single chip, it can become problematic as the PLLs may interfere
with each other, degrading the performance of the PLLs. Interference between PLLs
may depend on the physical distance between the PLLs and the frequency difference
between the PLLs.
[0005] Techniques are known for creating additional RF clock signals without adding a new
PLL. One such technique uses an up-converter to convert input modulation clock signals
and a local RF clock signal into the desired RF clock signal, where the frequency
of the desired RF clock signal,
fc, is equal to either
f0 + fm or
f0 - fm. For simplicity, here we only discuss the situation in which
fc=
f0 + cm, assuming that
fm can be either positive or negative. Here,
fo is the frequency of the local RF clock signal and
fm is the fundamental frequency of the input modulation clock signal. In practice, however,
the signal output from the up-converter is not an ideal signal, and instead of creating
a single tone clock signal spectrum, it may also generate undesired tones in its spectrum,
expressed as:

[0006] In this equation,
i , j, ni and nj are integers. The energy of the clock signal created is limited, so β
ij can become extremely small for large i and
j. Setting n
i and n
j to five would, in most cases, be sufficient to account for most significant tones.
When
i=1 and
j=1, the tone for the created clock signal, the desired tone, f
c=f
0 +f
m, is generated. Similarly, when
i=1 and
j=-1, the tone for the created clock signal, the desired tone, f
c=f
0 -f
m, is generated. The other created tones are spurious tones (i.e., unwanted tones).
The most harmful of these tones may be the tones that are close to the created clock
tone. For example, when
i=1 and
j=1 is the desired clock tone, the tones created at
i=1 and
j={-5,-4,-3,-2,-1,0,2,3,4,5} are harmful. Here, these tones are referred to as modulation
tones, as they are created by the non-linearity in handling the modulation signal.
Other tones, created when
i≠1, are referred to as harmonic tones, and they are normally located at the centers
or neighborhood of the harmonics ofRF clock
f0.
[0007] When generating additional RF clock signals using an up-converter, it is desirable
to eliminate the undesirable modulation and harmonic tones as much as possible.
SUMMARY
[0008] It is an object of the invention to overcome at least some of the above disadvantages,
and to provide for clock generation circuits that can generate a number of high quality
RF clock signals.
[0009] An embodiment described herein may provide for a clock generation circuit to generate
a target clock signal. The clock generation circuit may include a pattern generator
to generate a pre-distorted version of a modulation signal from patterns stored by
the pattern generator. An up-converter may up-convert the pre-distorted version of
the modulation signal using a radio frequency local oscillator signal to obtain an
RF clock signal having a desired frequency tone. A tone detection circuit may receive
the RF clock signal and detect amplitudes and phases corresponding to a presence of
unwanted tones. A controller may generate pre-distortion data based on the detected
amplitudes and phases and control the pattern generator to generate the pre-distorted
version of the modulation signal based on the pre-distortion data.
[0010] In some implementations, a notch filter may be implemented to filter the RF clock
signal, where the notch filter is constructed to suppress a third order harmonic tone
of the RF clock signal, for reducing the inter-modulation introduced by following
digital gate, such as clock driver or divider.
[0011] In one possible embodiment, the RF clock signal or filtered RF clock signal may be
down-converted to low frequency and sampled by an ADC or a sample-and-hold circuit
to detect the unwanted tones.
[0012] Another possible embodiment described herein may provide a method of producing an
RF clock signal. The method may include mixing a pre-distorted modulation signal and
a radio frequency local oscillator signal, to obtain the RF clock signal having a
desired frequency tone. Unwanted, spurious frequency tones in the RF clock signal
may be detected. Pre-distortion data, based on the detected unwanted frequency tones,
may be generated. Data patterns, based on the pre-distorted data, may be written to
a memory, data patterns, when read out from the memory, may include amplitude and
phase values designed to cancel the unwanted frequency tones in the RF clock signal.
[0013] Another embodiment described herein may include a mobile terminal that includes an
antenna; a processing system to control the operation of the mobile terminal; and
a transceiver component, connected to the processing system and the antenna, to wirelessly
send and receive data. The transceiver component may include a clock generation circuit
for generating an RF clock signal. The clock generation circuit may include an up-converter
to up-convert a pre-distorted modulation signal using a radio frequency local oscillator
signal to obtain the RF clock signal having a desired frequency tone. A pattern generator
may generate the pre-distorted modulation signal to cancel effects of non-linearities
of the up-converter. A tone detection circuit may receive the RF clock signal and
detect unwanted tones in the RF clock signal. A controller may control the pattern
generator to generate the pre-distortions in the modulation signal based on the detected
unwanted tones.
[0014] In some implementations, the pattern generator may be implemented using a number
of random access memories (RAMs), each of the random access memories storing a pattern
corresponding to a tone. An adder may digitally sum outputs of each of the random
access memories to generate the pre-distorted version of the modulation signal.
[0015] In some implementations, the pattern generator may include a first random access
memory to store the patterns, each of the patterns corresponding to one of the unwanted
modulation tones; a second random access memory to store pre-distortion data corresponding
to one of the tones; and an adder for digitally summing outputs from the first random
access memory and the second random access memory, where at least some of the outputs
of the adder are written back to the first random access memory for updating the patterns.
[0016] In general, the pattern generator may include a pair of pattern generators to respectively
generate the pre-distorted version of the modulation signals for an in-phase channel
and quadrature channel. In each channel, a digital-to-analog converter may receive
a pre-distorted version of a modulation signal from the pattern generator and output
an analog version of the pre-distorted modulation signal. A low pass filter may receive
the analog version of the pre-distorted modulation signal and output a low pass filtered
signal to the up-converter.
[0017] In some implementations, the controller of the clock generation circuit may control
the tone detection circuit to detect the amplitudes and phases during initial calibration
or operation mode test of the clock generation circuit. The controller of the clock
generation circuit may control the tone detection circuit to sequentially detect the
presence of a number of the unwanted tones in the RF clock signal or the filtered
version of the RF clock signal, and modify the pre-distorted pattern stored in random
access memory, to meet spectrum mask requirements.
[0018] In some implementations, the tone detection circuit of the clock generation circuit
may include a first mixer to receive the RF clock signal or a filtered version of
the RF clock signal as an input signal and down-convert the input signal based on
a radio frequency local oscillator signal to create a baseband representation of the
unwanted tones; and a second mixer to further down convert one of the baseband tones
into a DC signal. From in-phase (I) and quadrature phase (Q) channels, the amplitude
and phase of one of the down-converted unwanted tones may be sampled as residual error
signals corresponding to the unwanted tones. Based on the residual error signals,
the controller may modify the pre-distorted pattern to minimize the amplitudes of
the unwanted tones.
[0019] In some implementations, a combiner may combine the down-converted RF clock signal
with the modulation signal to remove the down-converted wanted tone in the RF clock
signal.
[0020] In some implementations, the pattern generator may include a random access memory
to store patterns corresponding to the modulation signal, where the stored patterns
include at least one stored pattern for each unwanted tone detected by the tone detection
circuit.
[0021] In some implementations, the controller may control the tone detection circuit to
detect the amplitudes and phases of the unwanted tones during initial calibration
or operation mode test of the clock generation circuit.
[0022] In some implementations, the controller may control the tone detection circuit to
sequentially detect the presence of a plurality of the unwanted tones in the RF clock
signal.
[0023] In some implementations, the tone detection circuit may include a first mixer to
receive the RF clock signal as an input signal and also use the RF clock signal as
the radio frequency local oscillator signal, thereby forming a self-mixing, to down
convert the RF clock signal..
[0024] In some implementations, the tone detection circuit may include a first mixer to
down-convert the RF clock signal, a sample-and-hold circuit to sub-sample the down-converted
RF clock signal, and an analog-to-digital converter to convert the sampled signal
to generate the pre-distortion data.
[0025] In some implementations, the tone detection circuit may include in-phase and quadrature
channel, each comprising a first mixer to down-convert the RF clock signal, a sample-and-hold
circuit to sub-sample the down-converted RF clock signal, and an analog-to-digital
converter to convert the sampled signal to generate the pre-distortion data.
[0026] Systems and/or methods described herein may provide for the generation of a clock
signal, such as an RF clock signal, using an input RF local oscillator signal and
a lower frequency modulation signal. Multiple different clock signals may be generated
based on the RF local oscillator signal in which tone cancellation is applied to the
generated clock signals to clean unwanted tones in the clock signals.
BRIEF DESCRIPTION OF THE DRAWINGS
[0027]
Fig. 1 is a diagram of exemplary device where an RF clock generation circuit and method
described herein may be implemented;
Fig. 2 is a diagram illustrating an exemplary implementation of an RF clock circuit;
Fig. 3 is a diagram illustrating an exemplary spectrum of an up-converter;
Fig. 4 is a diagram illustrating an exemplary spectrum of a clipped clock signal;
Fig. 5 is a diagram illustrating an exemplary implementation of a digital pattern
generator;
Fig. 6 is a diagram illustrating another exemplary implementation of a digital pattern
generator;
Fig. 7 is a diagram illustrating an exemplary implementation of an up-converter;
Fig. 8 is a diagram illustrating an exemplary implementation of a filter;
Fig. 9 is a diagram illustrating an exemplary implementation of a tone detection circuit;
Fig. 10 is a diagram illustrating exemplary frequency spectrums of various signals
in a tone detection circuit;
Fig. 11 is a diagram of an implementation of a tone detection circuit using sample
and hold circuits and a single mixing stage;
Fig. 12 is a diagram illustrating a timing relationship between a clock signal used
to trigger sample and hold circuits and a modulation signal; and
Fig. 13 is a flow chart illustrating an exemplary process for performing tone cancellation.
DETAILED DESCRIPTION
[0028] The following detailed description refers to the accompanying drawings. The same
reference numbers in different drawings may identify the same or similar elements.
Also, the following detailed description does not limit the invention.
[0029] Fig. 1 illustrates a diagram of an exemplary device 110, such as user equipment (UE),
in which an RF clock signal generator may be implemented. As shown in Fig. 1, UE 110
may include an antenna 115, transceivers (TX/RX) 120, a processing system 130, and
a user interface system 140. Device 110 may particularly be, for example, a mobile
terminal in a wireless communication network.
[0030] Antennas 110 may include one or more omni-directional antennas. Transceivers 120
may be associated with antennas 115 and may include transceiver circuitry for wirelessly
transmitting and/or receiving data with a base station in the wireless network. Transceivers
120 may particularly include one or more RF clock signal generator circuits 125 that
each generate a periodic RF clock signal. Each of RF clock signal generator circuits
125 may generate its clock signal based on a single set of local quadrature RF clock
signals, such as generated using a PLL, and a low frequency modulation signal. RF
clock signal generator circuits 125 may share the input set of local quadrature RF
clock signals, thus allowing a number of RF clock signals to be generated without
each signal requiring a PLL. Implementations of RF clock signal generator circuits
125 will be described in more detail below.
[0031] Processing system 130 may control the operation ofUE 110. Processing system 130 may
process information received via transceiver 120 and/or user interface system 140.
As illustrated in Fig. 1, processing system 130 may include a processing unit 132
and a memory 134. Processing unit 132 may include one or more processors, microprocessors,
application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs),
or the like. However, systems and/or methods described herein are not limited to any
specific combination of hardware circuitry and software.
[0032] Although Fig. 1 shows exemplary components ofUE 110, in other embodiments, UE 110
may contain fewer, different, differently arranged, or additional components than
depicted in Fig. 1. Alternatively or additionally, one or more components ofUE 110
may perform one or more other tasks described as being performed by one or more other
components ofUE 110.
RF CLOCK CIRCUIT
[0033] Fig. 2 is a diagram illustrating an exemplary implementation of one ofRF clock signal
generator circuits 125. In general, an RF clock signal generator circuit 125 may use
an up-converter to mix a low frequency modulation signal with a set of local quadrature
RF clock signals to obtain an RF output clock signal. Tone cancellation is performed
to reduce spurious tones in the output signal to obtain a high quality RF clock signal.
[0034] As particularly shown, RF clock signal generator circuit 125 may include two parallel
connected channels, one for In-phase (I) channel and another for the quadrature (Q)
channel. Each channel may include a digital pattern generator (DPG) 205, a digital-to-analog
(DAC) converter 210, and a low pass filter (LPF) 215. RF clock signal generator circuit
125 also includes an up-converter (mixer, MIX) 220, a filter 225, and a clock driver
or a clock divider (CD) 230. Additionally, feedback information from filter 225 may
be received by tone detection circuit 235, the output of which may be received by
controller 240. Controller 240 may control DPGs 205 based on the feedback to, for
example, pre-distort the input modulation signal to compensate for nonlinearities
of up-converter 220.
[0035] RF clock signal generator circuit 125 may generate the output RF clock signal,
ck(t), having a desired frequency
fc, based on the low frequency input modulation signal having a fundamental frequency,
and the local quadrature RF clock signals 250 (
fo). Local quadrature RF clock signals 250 may be generated using a PLL and may be input
to a number of similarly constructed RF clock generator circuits 125 for generating
a corresponding number of different RF output signals.
[0036] DPGs 205 may include a number of random access memories (RAMs), the accessing of
which may be controlled by controller 240 and a low-frequency modulation signal
mt with a fundamental frequency
fm (or an integer multiple, K, of f
m). The digital modulation signal
mt is created by DPGs 205 illustrated in Fig. 2 and may be directly read out at clock
frequency
fro =K
fm, where K is an oversampling factor. DPGs 205 may generate digital modulation signals
mt for I and Q channels, which may represent digital patterns generated based on lookup
operations into RAMs of DPGs 205. Because the I and Q channel patterns have phase
shift of 90 degrees, a smart hardware arrangement for DPGs 205 can save hardware cost.
The lookup operations may generate sinusoid signals with a desired amount of pre-distortion.
Exemplary implementations of DPGs 205 will be described in more detail below.
[0037] DACs 210 may include two DACs, one for each of the I and Q channels output by DPGs
205. DACs 210 may operate to convert the digital signals into analog signals.
[0038] LPFs 215 may remove high frequency components from the input signals and output quadrature
I(t) and Q(t) channels that are to be mixed with the local quadrature RF clock signals
250 that are also input to up-converter 220.
[0039] Up-converter 220 mixes the input clock signals and modulation signals to generate
an output signal having a desired tone at (
fo +
fm) or
(fo - fm) (where
fo refers to a local oscillator RF signal). In practice, because of the non-idealities
that the load impedance of up-converter 220 is not zero and the output impedance of
the active components are not infinite, up-converter 220 normally behaves as a nonlinear
transfer function that introduces unwanted harmonic tones and modulation tones. As
a comparison, an ideal up-converter may convert the input modulation signals and the
local quadrature RF clock signals 250 into the wanted clock signal based on the following
equation:

[0040] The output signal in equation (1) contains only a single tone, the wanted clock signal.
Here, ω
0=2
πf
0, ω
s=2
πf
s and ω
c=2
πf
c, respectively.
[0041] Up-converter 220, as previously mentioned, may also create unwanted harmonic tones.
The harmonic tones may be created at frequencies of
nfo±
mfm, where n and m are integers. The case when n=3 and m=1 may create a particularly strong
3
rd order harmonic tone, f
h3, at the output of up-converter 220. Note also that harmonic tones at
nfc = nfo±
nfm should not introduce problems because, as in a periodical pulse series, it has the
fundamental frequency
fc and harmonics as well.
[0042] Fig. 3 is a diagram illustrating an exemplary spectrum of a particular up-converter,
such as up-converter 220. As shown in Fig. 3,
where f0=2GHz
,fm=50MHz and
fc=2.05GHz, the harmonic tone, f
h3=3
f0- fm =5.95GHz , labeled 305 is about 40 dB lower than the created clock tone 310 at
fc =2.05GHz. When a clock signal with a spectrum such as shown in Fig. 3 is fed into
a digital gate, such as a clock driver CD 230, the harmonic tones may be mixed by
the clock driver CD 230 as it is a strong nonlinear block, which creates unwanted
tones which may also include harmful modulation tones.
[0043] Fig. 4 is a diagram illustrating an exemplary spectrum of a clipped clock signal
performed by a digital gate. The harmonic tone, f
h3, is weaker than that in Fig. 3. However, as shown, part of the power is down-converted
into the harmful modulation tones close to the created clock, substantially increasing
the amplitudes of the unwanted modulation tones.
[0044] Referring back to Fig. 2, filter 225 may include a bandpass filter and a notch filter.
Filter 225 can particularly operate to suppress the 3
rd order harmonic tone before it is clipped by the digital gate CD 230. One possible
implementation of filter 225 will be described in more detail below.
[0045] CD 230 converts its analog input signal into the desired RF digital clock signal,
ck(t), having frequency
fc. CD 230 can also be a clock buffer or a clock divider or any other kind of digital
gate. Clock driver and divider circuits are known and will not be described further
herein.
[0046] Tone detection circuit 235 may receive the output of filter 225 and detect residual
error information for one of the unwanted tones in the signal
so(t), output from filter 225. Tone detection circuit 225 may, at different times, detect
residual error information for different ones of the unwanted tones. Exemplary implementations
of tone detection circuit 235 will be described in more detail below.
[0047] Controller 240 receives the residual error information from tone detection circuit
235 and, based on the residual error information, generates control signals to DPGs
205, which output the pre-distorted sinusoids used to cancel the unwanted tones. Controller
240 may be implemented in, for example, a digital signal processor (DSP), application
specific integrated circuit (ASIC), or other hardware or hardware/software based component.
The operation of controller 240 will be described in more detail below.
[0048] Although Fig. 2 shows exemplary components of an RF clock generator circuit 125,
in other embodiments, RF clock generator circuit 125 may contain fewer, different,
differently arranged, or additional components than depicted in Fig. 2.
[0049] Implementations of DPGs 205, up-converter 220, filter 225, and tone detection circuit
235 will now be discussed in additional detail with reference to Figs. 5-13.
[0050] Fig. 5 is a diagram illustrating an exemplary implementation of a DPG 205 for one
channel. The second one of DPG 205 for another channel may be similarly implemented.
As previously mentioned, each of DPGs 205 may operate to generate a pre-distorted
modulation signal in which the distortion is controlled to reduce the nonlinearity
of up-converter 220 and thereby suppress unwanted modulation tones.
[0051] As shown in Fig. 5, DPG 205 may include random access memories (RAMs) 505-1 through
505-k, the output of each of which may be connected to a corresponding buffer register
510-1 through 510-k. The output of registers 510 are input to adder 520. Adder 520
digitally sums its input value and output the sum result.
[0052] Each of RAMs 505 can be modified from an initial pattern representing the sinusoid
modulation signals to updated patterns, that, when read out, as controlled by controller
240, represent pre-distorted sinusoid signals including tuned harmonics that are to
be added to the sinusoid signal(s). The particular pre-distorted sinusoid signals
at the output of a RAM 705 may be controlled by controller 240 based on the vector
signals Din and Ctr. Din, for example, may represent data streams through which controller
240 may program RAMs 505 to include fundamental sinusoid signals and the pre-distortion
signals, i.e, the harmonics of the fundamental sinusoid signals with proper phase
and amplitude. Ctr may represent address lines through which controller 240 may select
the values from RAMs to generate the desired pre-distorted signals or store the data
stream Din into a selected RAM.
[0053] Registers 510 may receive the outputs of RAMs 505 and buffer the outputs before forwarding
the outputs to adder 520. The clock signal Ckf
m may be generated by controller 240 and used to trigger the latching of data from
registers 510. In general, Ckf
m is at a frequency kf
m, where k is an integer over-sampling ratio, which may be much larger than 1. Larger
over-sampling ratio values k may result in greater accuracy to reach the desired phase
alignment, especially for high order modulation cancellation tones. However, using
large k values means increasing the frequency for the digital circuits. To get good
phase alignment for higher order tones and maintain lower over-sampling ratios, one
technique may include introducing several different tunable delay elements in the
clock distribution for Ckfm.
[0054] In one implementation of RAMs 505, k may be five, i.e., DPG 205 may include five
RAMs 505: RAM1 through RAM5. In this case, RAM1 may store the fundamental sinusoid
signals, RAM2 may store a pre-distortion signal for eliminating local oscillator (LO)
leakage, RAM3 may store a pre-distortion signal for eliminating 2
nd order distortion due to device mismatch, RAM4 may store a pre-distortion signal for
eliminating 3
rd order distortion, and RAM5 may store a pre-distortion signal for eliminating 5
th order distortion.
[0056] Fig. 6 is a diagram illustrating a second possible exemplary implementation of one
channel of DPGs 205. The second channel of DPGs 205 may be similarly implemented.
[0057] As shown in Fig. 6, DPG 205 may include random access memories RAMm 605 and RAMsm1610,
address generators 615 and 620, registers 625 and 630, and an adder 640. RAMm 605
is the "main" memory in which patterns are stored and RAMsm1610 is a smaller RAM that
may receive data from controller 240 (Din) during operation (i.e., during modulation
tone cancellation). RAMm 605 may generally correspond to a merged version of RAMs
505 (Fig. 5). Address generators 615 and 620 provide addresses to RAMm 605 and RAMsm1
610, respectively. Registers REGm 625 and REGsm1630 may connect to the outputs of
RAMm 605 and RAMsm1610, respectively, to buffer the output values before the output
values are summed by adder 640.
[0058] As shown, the input ports of RAMm 605 can be connected to receive write data, Din,
from both controller 240 and the output, Dout, of adder 640, according to a configuration
controlled by controller 240.
[0059] In operation, controller 240 may program RAMsm1 610 to "tune" the amplitude and angle
of the signal output from adder 640 so that one unwanted modulation tone in the RF
clock signal is minimized. The output of adder 640 may then be sent to RAMm 605 to
update the previous contents in RAMm 605. In this manner, unwanted modulation tones
can be cancelled one-by-one, under the control of controller 240.
[0060] In some implementations, in RAMsm1 610, the read-out initial address can be set to
an offset, instead of zero to correspond to a time delay or a phase offset. Changing
the initial offset address may lead to an initial phase change that can be used to
tune the phase during calibration. In some implementations, a digital multiplier can
also be inserted between register REGsml 630 and the RAMs 610, potentially allowing
for the performance of amplitude tuning during an amplitude calibration operation.
Phase calibration and amplitude calibration may be needed to eliminate unwanted tones.
Phase calibration may be performed first, followed by amplitude calibration. Register
REGsml may be triggered at a delayed clock phase of the clock signal Ckfm, with insertion
of a variable delay element in the clock path (not shown).
[0061] Although Fig. 5 and 6 show exemplary components of a DPG 205, in other embodiments,
DPG 205 may contain fewer, different, differently arranged, or additional components
than depicted in Fig. 5 and 6.
[0062] Fig. 7 is a diagram illustrating an exemplary implementation of up-converter 220.
Up-converter 220 may be an up-converter based on the known Gilbert mixer topology.
Up-converter 220 may include a double balanced mixer topology, which may give good
cancellation of even order distortion and LO leakage.
[0063] As particularly shown in Fig. 7, up-converter 220 may include a double balanced mixer
710 and switches 730, which may be used to control the desired output frequency for
the generated RF clock signal.
[0064] Double balanced mixer 710 may perform the up-conversion of the input signals. Double
balanced mixer 710 may include a first section 715 and a second section 720. Each
section 715 and 720 may include eight transistors, connected as shown. The output
of first section 715 is labeled as nodes O
ip and O
in. The output of second section 720 is labeled as nodes O
qp and O
qn.
[0065] Switches 730 may select, based on the value of input pin P
n, how the nodes O
ip, O
in, O
qp, and O
qn are selected for generating output at nodes v
op and v
on. For example, when P
n is logic high, switches 730 may connect node v
op to nodes O
ip and O
qn, and node v
on to nodes O
in and O
qp. In this case, the output of up-converter 220, after processing by filter 225, may
generate a wanted RF clock signal of frequency at f
o+f
m. When P
n is logic low, switches 730 may connect node v
op to nodes O
ip and O
qp, and node v
on to nodes O
in and O
qn. Controller 240 may control the value of P
n. In this case, the output of up-converter 220, after processing by filter 225, may
generate a wanted RF clock signal of frequency at f
o-f
m. Control signal E may act as an enable signal, that, depending on its value, either
enables or disables up-converter 220.
[0066] Fig. 8 is a diagram illustrating an exemplary implementation of filter 225. Filter
225 may generally operate to suppress the 3
rd order harmonic tone before digital clipping by the clock driver or other kind of
digital gate CD 330. Filter 225 may implement a band pass filter and a notch filter.
[0067] Filter 225 may particularly include capacitor 810, transformer 815, capacitor 820,
inductor 825, and capacitor 830. Capacitors 810, 820, and 830 may be tunable capacitors
that can be tuned by controller 240.
[0068] In one implementation, capacitor 810 may be tuned so that capacitor 810 and primary
coil of transformer 815 are at the target clock frequency,
fc, which may yield a maximum gain at
fc for the band pass filter. Capacitors 820 and 830, and inductor 825 may be tuned separately
to the 3rd order harmonic frequency. An alternative possible technique for tuning
filter 225 may include actively tuning capacitor 810 to
fc and passively tuning capacitors 820 and 830 based on the tuning of capacitor 810.
For example, the same control word from controller 240 may be used to tune capacitors
810, 820, and 830.
[0069] Fig. 9 is a diagram illustrating an exemplary implementation of tone detection circuit
235. In general, tone detection circuit 235 may sequentially (i.e., one at a time)
detect unwanted tones. The detected tone information is input to controller 240, which
uses the information relating to the unwanted tones to control DPGs 205 to pre-distort
the modulation signal.
[0070] Tone detection circuit 235, as shown in Fig. 9, includes a series connected first
mixer 905, adder 910, first low pass filter (LPF1) 915, second mixer 920, second low
pass filter (LPF2) 925, amplifier (AMP) 930, and analog-to-digital converter (ADC)
935. First local oscillator clock signals at
frequency f0 may be input to first mixer 905 via a first switch (SW1) 940. A variable signal at
a selectable frequency ofn-times the modulation signal frequency (
nfm, where
n is an integer) may be input to second mixer 920 via a second switch (SW2) 945.
[0071] In operation, first switch 940 may receive local oscillator signals
cos(ωot) and
sin(ωot), at the same frequency
fo as the local quadrature RF clock signals, and output one of the signals based on
a switch selection signal
cs1. First mixer 905 down converts the input signal based on the local oscillator signal
from first switch 940 to down-converted signal L(t). Adder 910 may then add a fundamental
modulation signal to L(t), to remove the down-converted wanted tone in the RF clock
signal and generate signal E1(t). First low pass filter 915 may remove high frequency
harmonics from E 1 (t). The signal output from first low pass filter 915 may be further
down converted into error signal E2(t) by second mixer 920. Signal E2(t) may represent
a baseband version, around DC, of the particular unwanted tone that is being detected.
Second mixer 920 may receive the output of first low pass filter 915 and the output
of second switch 945. Second switch 945 may output, based on selection signal
cs2, either
cos(nωmt) or
sin(nωmt). The signals input to second switch 945 are low frequency clock signals having frequencies
at the harmonics of the modulation clock frequency, i.e.,
nfm, where
n is an integer (e.g., n = 0, 1, 2, 3, 4, 5). The value
of n may be sequentially varied by controller 240 to perform tone detection for different
modulation clock frequency harmonics.
[0072] Second low pass filter 925 may remove ripples in signal E2(t) to create a clean signal
E(t). The DC component ofE(t) may represent the residual error signal. Amplifier 930
may boost the residual error signal and analog-to-digital converter 935 may convert
the residual error signal to digital output data, e(i), which is sampled and quantized
at discrete times based on sampling clock signal
ck.
[0073] The two input signals to second switch 945 may be selected based on which of the
quadrature components of the tone is to be detected. Controller 240 may control second
switch 945 to select the input clock signal for second mixer 920 in an interleaved
manner. Controller 240 may calculate the amplitude of the interleaved quadrature residual
signals
ei(i) and
eq(i) as

[0074] The phase of the residual error may be calculated as

[0075] The residual error amplitude and phase can be used to estimate the required pre-distortion
amplitude and phase.
[0076] First switch 940 may be used in a similar manner for the local oscillator signal
input to first mixer 905. First switch 940 may be switched based on the component
of the unwanted tone that is being detected so the tone detection can work in an interleaved
mode for detecting in-phase and quadrature phase components of the unwanted tone.
In second switch 945, the parameter
n can be set for a particular tone. For example,
n is set to zero in order to detect phase and amplitude of the local oscillator leakage.
[0077] Fig. 10 is a diagram illustrating exemplary frequency spectrums of various signals
in tone detection circuit 235. In particular, from top to bottom, the following frequency
spectrums are shown: (1) the frequency spectrum, SO(f), of the generated RF clock
signal,
so(t), the input to tone detection circuit 235; (2) the frequency spectrum of the local
oscillator clock; (3) the frequency spectrum, L(f), of signal L(t); and the frequency
spectrum, E1(f), of signal E1(t).
[0078] The frequency spectrum SO(f) of the input signal includes the desired tone, i.e,
the generated RF clock signal, T1, and a number of unwanted tones (T0, T2, T3, and
T5). Tone T0 for instance, may be created by local oscillator leakage or device mismatch.
Tones T3 and T5, on the other hand, may be created by the odd orders of non-linearity
of up-converter 220. Tone detection circuit 235 may be controlled by controller 240
to isolate, one at a time, tones T0 (n=0), T2 (n=2), T3 (n=3), and T5 (n=5). In this
manner, error information corresponding to each of these unwanted tones may be generated
and output to controller 240.
[0079] In frequency spectrum L(f), it can be seen that the wanted tone, T1, has a relatively
large amplitude compared to the unwanted tones. The large amplitude of T1 may drive
the circuits after first mixer 905 into non-linear operation. In order to avoid this,
a tone at the same amplitude but opposite in phase may be injected by adder 910 to
suppress this tone. The output frequency spectrum of adder 910, tone E1 (f), illustrates
suppression of tone T1. At this point, the signal can be further mixed by second mixer
920 to down-convert the chosen unwanted signal before low pass filtering to isolate
the desired unwanted signal before it is amplified and digitally sampled.
[0080] In the exemplary implementation of tone detection circuit 235 in Figs. 9 and 10,
the local oscillator clock signal is used in first mixer 905 and T1 is down-converted
into the modulation frequency. The large tone, T1, is then removed in order to keep
the circuit in a linear region.
[0081] In an alternative technique for implementing tone detection circuit 235, the generated
clock signal may instead be self-mixed and down-converted to a DC signal, then detected
in amplitude. As the signal is amplitude detected, T 1 may then be removed by high
pass (or band pass) filtering. In this situation, the amplitude of unwanted modulation
tones can then be found either by directly feeding the signal to AD converters with
sufficient bandwidth, or by feeding the signal to a chain similar to second mixer
920, low pass filter 925, amplifier 930, and analog-to-digital converter 935.
[0082] In another alternative technique for implementing tone detection circuit 235, instead
of using a high speed ADC 935 (Fig. 9), high speed sample and hold circuits and a
low speed ADC may instead be used. The sample and hold circuits are possible because
the residual error signals are periodic.
[0083] Fig. 11 is a diagram of an implementation of tone detection circuit 235 using sample
and hold circuits and a low speed ADC, in which only a single mixing stage is used.
As shown, the input signal is provided through an upper circuit path 1100 and a lower
circuit path 1150. Upper circuit path 1100 may include a mixer 1105, a low pass filter
1110 (LPF), a sample and hold circuit (SH) 1115, an optional low pass filter (LPF),
an amplifier 1125, and a low speed analog-to-digital converter (ADC) 1130. Lower circuit
path 1150 may similarly include a mixer 1155, a low pass filter 1160 (LPF), a sample
and hold circuit (SH) 1165, an optional low pass filter (LPF) 1170, an amplifier 1175,
and a low speed analog-to-digital converter (ADC) 1180. The input signal may be split
between upper circuit path 1100 and lower circuit path 1150. In upper circuit path
1100, the input signal may be down-converted by mixer 1105, and then input to sample
and hold circuit 1115. Sample and hold circuit 1115 may sample its input based on
the clock signal
cks. The clock signal
cks may have a period that is close to a multiple (
n) of the period of the modulation frequency,
fm. The output of sample and hold circuit 1115 may be filtered by low pass filter 1120,
amplified by amplifier 1125, and converted to a digital representation by analog-to-digital
converter 1130. Lower circuit path 1150 functions similarly to upper circuit path
1100, except that the signal input to mixer 1155 is in quadrature phase with respect
to the signal input to mixer 1105.
[0084] Although Fig. 9 and 11 show exemplary components of tone detection circuit 235, in
other embodiments, tone detection circuit 235 may contain fewer, different, differently
arranged, or additional components than depicted in Figs. 9 and 11. For the two-chain
structure of Fig. 11, it is also possible to merge the two chains together using a
time interleaving method.
[0085] Fig. 12 is a diagram illustrating a timing relationship between the clock signal,
cks, used to trigger sample and hold circuits 1115 and 1165, and the modulation signal
at frequency,
fm. The relationship between
fm and
cks may be defined by

or,

where
fs is the frequency of
cks and
TS is the period of
cks, and where
m and
j are integers. When
m>1, the sampling frequency is lower than f
m, so that sub-sampling is performed. When
m=0, oversampling is performed, where
fs is
j times higher than
fm. Here
j is the number of sampling points and may equal the product of all order numbers of
the unwanted tones to detect. As such, sample-and-hold circuits 1115 and 1165 may
sample the clock residual error signals every
m periods. In addition, a skew of ±1/
jTs, may be introduced. For sampling of one complete period,
j *m+1 periods at frequency
fm may be needed.
[0086] With sub-sampling, analog-to-digital converters 1130 and 1180 may run at relatively
low frequencies, reducing both power consumption and design requirements.
[0087] Controller 240 may generally operate to control DPGs 205 and tone detection circuit
235 to perform tone cancellation of unwanted signals in the output clock signal. Controller
240 may, for example, write data patterns to DPGs 205 so that DPGs 205 generate the
distorted sinusoid signals designed to cancel the nonlinearities of up-converter 220.
Controller 240 may also receive the digital residual error signals from tone detection
circuit 235, and produce the pre-distorted error signals and modify or generate the
data patterns for DPGs 205. Generation of the data patterns for DPGs 205 may be performed,
for example, at calibration time or during normal operation.
[0088] Fig. 13 is a flow chart illustrating an exemplary process 1300 for performing tone
cancellation by controller 240.
[0089] Process 1300 may include minimizing the amplitude of the residual error signal due
to an unwanted tone for the desired RF clock signal (block 1310). The desired RF clock
signal may correspond to the tone at
n=1. In tone detection circuit 235, as shown in Fig. 9, for instance, controller 240
may set
n =
1 for the sinusoid signals input to mixer 920. Controller 240 may then adjust the weight
coefficient (A
1) and the output phase (Φ
1), which are used in the generated tone input to adder 910 (Fig. 9), so that the amplitude
of the error signal is minimized. More particularly, controller 240 may minimize

[0090] Process 1300 may also include setting
n equal to the value of a tone that is to be cancelled (block 1320). As previously
discussed, the tones that may need to be cancelled are the tones in the set
n={3, 5, 0, 2}. For example,
n may be set to three. Tone detection circuit will then detect the tone at f
c-4f
m. Controller 240 may then tune the weight coefficient (A
1) and the output phase (Φ
1), which are used in the generated tone input to adder 910 (Fig. 9), so that the amplitude
of the error signal is minimized (block 1330). The data pattern for the signal having
the determined weight coefficient and output phase may be written to DPGs 205 (block
1340).
[0091] The process shown in blocks 1320-1340 may be repeated for each tone that is to be
cancelled, i.e., for each additional value of
n (block 1350).
[0092] The cancellation results for process 1300 may depend on the sensitivity of tone detection
circuit 135. Because the wanted clock signal is a deterministic signal, the output
signal can be a DC signal and the sensitivity can be set to be relatively high. To
avoid flicker noise, chopping can be used in the amplifier and the analog-to-digital
converters. Another factor that can influence the tone cancellation is the accuracy
of the generated cancellation tones. In one implementation, before applying pre-distortion
the amplitude of the spurious tones were at least 40dB lower than that of the created
clock signal. Additional attenuation of 40dB is then typically sufficient.
[0093] The techniques described above can be used in carrier aggregation, both in the receiver
and transmitter. In carrier aggregation applications, several RF clock signals may
be required for the Rx and/or the Tx chain in UE 110. Using a clock generator with
tone cancellation may solve the issue of multiple RF clocks without degradation of
the performance in the Rx (or Tx) chain.
[0094] Systems and/or methods described herein may generate multiple RF clock signals without
needing a corresponding PLL for each clock signal. Instead, multiple high quality
clock signals are generated using a single PLL in which tone cancellation is applied
to the generated clock signals to clean unwanted tones in the clock signals. In the
tone cancellation, unwanted tones are detected and the modulation signal is pre-distorted
in a way designed to offset each of the unwanted tones. The tone cancellation can
be used to cancel both unwanted modulation tones due to non-linearity of an up-converter
and clock output leakage caused by device mismatch. Further, power consumption of
tone detection circuit 235 may be relatively low, as tone detection circuit 235 may
only need to be used occasionally, such as at a regular (small duty cycle) interval
or at initial calibration. Although RF clock signals are referred here, the method
and systems may be used to generate clock signals that have a desired frequency in
the microwave range.
[0095] The foregoing description of embodiments provides illustration and description, but
is not intended to be exhaustive or to limit the invention to the precise form disclosed.
Modifications and variations are possible in light of the above teachings or may be
acquired from practice of the invention. For example, while a series of blocks has
been described with regard to Fig. 13, the order of the blocks may be modified in
other embodiments. Further, non-dependent blocks may be performed in parallel. Even
though particular combinations of features are recited in the claims and/or disclosed
in the specification, these combinations are not intended to limit the invention.
In fact, many of these features may be combined in ways not specifically recited in
the claims and/or disclosed in the specification.
[0096] No element, block, or instruction used in the present application should be construed
as critical or essential to the invention unless explicitly described as such. Also,
as used herein, the article "a" is intended to include one or more items. Where only
one item is intended, the term "one" or similar language is used. Further, the phrase
"based on" is intended to mean "based, at least in part, on" unless explicitly stated
otherwise.
1. A clock generation circuit for generating an RF clock signal, the clock generation
circuit comprising:
a pattern generator to generate a pre-distorted version of a modulation signal from
digital patterns stored by the pattern generator;
an up-converter to mix the pre-distorted version of the modulation signal and a radio
frequency local oscillator signal to generate the RF clock signal having a wanted
tone corresponding to a desired frequency;
a tone detection circuit to receive the RF clock signal and to detect both amplitudes
and phases corresponding to a presence of unwanted tones in the RF clock signal; and
a controller to generate pre-distortion data based on the detected amplitudes and
phases and control the pattern generator to generate the pre-distorted version of
the modulation signal based on the pre-distortion data.
2. The clock generation circuit of claim 1, where the pattern generator additionally
includes:
a plurality of random access memories, each of the random access memories storing
a pattern, of the digital patterns, corresponding to a tone; and
an adder for digitally summing outputs of each of the plurality of random access memories
to generate the pre-distorted version of the modulation signal.
3. The clock generation circuit of claim 1, where the pattern generator additionally
includes:
a first random access memory to store the patterns, each of the patterns corresponding
to a tone of a plurality of tones;
a second random access memory to store the pre-distortion data corresponding to one
of the plurality of tones; and
an adder for digitally summing outputs from the first random access memory and the
second random access memory, where at least some of the outputs of the adder are written
back to the first random access memory.
4. The clock generation circuit of claim 1, where the pattern generator additionally
includes a pair of pattern generators to respectively generate an in-phase pre-distorted
version of the modulation signal and a quadrature pre-distorted version of the modulation
signal.
5. The clock generation circuit of claim 1, further comprising:
a digital-to-analog converter to receive the pre-distorted version of a modulation
signal from the pattern generator and to output an analog version of the pre-distorted
modulation signal; and
a low pass filter to receive the analog version of the pre-distorted modulation signal
and output a low pass filtered signal to the up-converter.
6. The clock generation circuit of claim 1, further comprising a notch filter to filter
the RF clock signal; and wherein the notch filter is constructed to suppress a third
order harmonic tone of the RF clock signal to prevent additional modulation tones
introduced by a following clock driver or clock divider.
7. The clock generation circuit of claim 6, where the notch filter is implemented in
combination with a band pass filter constructed to pass a frequency corresponding
to the desired frequency of the RF clock signal.
8. The clock generation circuit of claim 1, where the controller controls the tone detection
circuit to detect the amplitudes and phases during initial calibration or operation
mode test of the clock generation circuit.
9. The clock generation circuit of claim 6, where the controller controls the tone detection
circuit to sequentially detect the presence of a plurality of the unwanted tones in
the RF clock signal.
10. The clock generation circuit of claim 1, where the tone detection circuit includes:
a first mixer to receive the RF clock signal as an input signal and down-convert the
input signal based on the radio frequency local oscillator signal; and
a second mixer to convert the down converted input signal to a signal that includes
a baseband representation of one of the unwanted tones.
11. The clock generation circuit of claim 10, wherein the radio frequency local oscillator
signal connected to the first mixer is replaced with the generated RF clock signal,
thereby forming a self-mixing to convert the RF clock signal to baseband signal, which
is amplitude detected.
12. The clock generation circuit of claim 10, further comprising a combiner to combine
the down-converted RF clock signal with the modulation signal to remove the down-converted
wanted tone in the RF clock signal.
13. The clock generation circuit of claim 1, wherein the tone detection circuit comprises
a first mixer to down-convert the RF clock signal, a sample-and-hold circuit to sub-sample
the down-converted RF clock signal, and an analog-to-digital converter to convert
the sampled signal to generate the pre-distortion data.
14. The clock generation circuit of claim 1, wherein the tone detection circuit comprises
in-phase and quadrature channel, each comprising a first mixer to down-convert the
RF clock signal, a sample-and-hold circuit to sub-sample the down-converted RF clock
signal, and an analog-to-digital converter to convert the sampled signal to generate
the pre-distortion data.
15. The clock generation circuit of claim 1, where the clock generator is implemented
in an electronic device, a wireless communication terminal, or a transceiver in a
base station, in radio frequency spectrum or microwave frequency spectrum.
16. A method of producing an RF clock signal, the method comprising:
mixing a pre-distorted modulation signal and a radio frequency local oscillator signal
to obtain the RF clock signal with a desired frequency tone;
sequentially detecting a plurality of unwanted frequency tones in the RF clock signal;
generating pre-distortion data based on the detected plurality of unwanted frequency
tones; and
writing data patterns, based on the pre-distortion data, to a memory that is used
to output the pre-distorted modulation signal, the data patterns, when read out from
the memory, including amplitude and phase values designed to cancel the plurality
of unwanted frequency tones in the RF clock signal.
17. The method of claim 16, further comprising:
converting the pre-distorted modulation signal to an analog signal; and
filtering the converted analog signal using a low pass filter.
18. The method of claim 16, further comprising:
band pass filtering the RF clock signal to pass the frequency tone equal to the desired
frequency.
19. The method of claim 16, further comprising:
filtering the RF clock signal, using a notch filter, to reduce a presence of a third
order harmonic tone in the RF clock signal to prevent additional modulation tones
introduced by a following clock driver or clock divider.
20. The method of claim 16, wherein detecting a plurality of unwanted frequency tones
in the RF clock signal comprises:
down converting the RF clock signal;
sub-sampling the down-converted RF clock signal; and
converting the sub-sampled signal to an analog signal.
21. A mobile terminal comprising:
an antenna;
a processing system to control the operation of the mobile terminal; and
a transceiver component, connected to the processing system and the antenna, to wirelessly
send and receive data, the transceiver component including a clock generation circuit
for generating an RF clock signal, the clock generation circuit including
an up-converter to up-convert a pre-distorted modulation signal and a radio frequency
local oscillator signal to obtain the RF clock signal having a desired frequency tone;
a pattern generator to generate the pre-distorted modulation signal, the pre-distorted
modulation signal being generated to include pre-distortions to cancel effects of
non-linearities of the up-converter;
a tone detection circuit to receive the RF clock signal and to detect unwanted tones
in the RF clock signal; and
a controller to control the pattern generator to generate the pre-distortions in the
modulation signal based on the detected unwanted tones.